LEE, Circuit Judge

Robert W. Clore (argued) and Christopher A. Bandas, Bandas Law Firm P.C., Corpus Christi, Texas; Timothy R. Hanigan and Vaughn M. Greenwalt, Lang Hanigan & Carvalho LLP, Woodland Hills, California; for Objector-Appellant Christine Knott.

Sam A. Miorelli (argued), Law Office of Sam Miorelli P.A., Orlando, Florida, for Objector-Appellant/Cross-Appellee Jan L. Miorelli, Personal Representative of the Estate of George P. Liacopoulos.

Christopher T. Cain, Scott & Cain, Knoxville, Tennessee, for Objector-Appellant W. Allen McDonald.

Steven A. Schwartz (argued) and Timothy N. Mathews, Chimicles Schwartz Kriner & Donaldson-Smith LLP, Haverford, Pennsylvania; Charles S. Fax and Liesel J. Schopler, Rifkin Weiner Livingston LLC, Bethesda, Maryland; Jeffrey M. Cohon, Law Offices of Jeffrey M. Cohon APC, Los Angeles, California; Jonathan D. Selbin and Andrew Kaufman, Lieff Cabraser Heimann & Bernstein LLP, New York, New York; David H. Weinstein and Robert Kitchenoff, Weinstein Kitchenoff & Sher LLC, Philadelphia, Pennsylvania; for Plaintiffs-Appellees.

Andrew J. Pincus (argued), Mayer Brown LLP, Washington, D.C.; Michael T. Williams, Allison R. McLaughlin, and Galen D. Bellamy, Wheeler Trigg O'Donnell LLP, Denver, Colorado; Stephen M. Shapiro, Timothy S. Bishop, Joshua D. Yount, and Chad M. Clamage, Mayer Brown LLP, Chicago, Illinois; for Defendants-Appellees/Cross-Appellants.

Oramel H. Skinner III (argued) and Dana R. Vogel, Assistant Attorneys General; Paul N. Watkins, Civil Litigation Division Chief; Mark Brnovich, Attorney General; Office of the Attorney General, Phoenix, Arizona; for Amicus Curiae Nine State Attorneys General.

Before: Richard R. Clifton and Kenneth K. Lee, Circuit Judges, and Frederic Block, District Judge.

LEE, Circuit Judge: Is it reasonable to award $14.8 million in attorney's fees in a class action settlement that provides $116.7 million in benefits to class members? But what if the class settlement is in fact worth only $4.2 million? We face these two dramatically divergent scenarios in large part because the settlement here offers "coupons" that may provide phantom benefits to most class members.

The parties settled a long-running class action lawsuit about malfunctioning "electronic control boards" in Whirlpool dishwashers. That settlement provided, among other things, coupons that consumers could use to buy a new Whirlpool dishwasher. The parties, however, could not agree on the value of this settlement, or the amount of attorney's fees for the plaintiffs’ counsel.

The district court approved the class settlement and awarded $14.8 million in attorney's fees based on a lodestar calculation of billable hours expended. We affirm the district court's approval of the settlement. But we vacate and remand the fee award because the district court erred in applying a lodestar-only methodology for the coupon portion of the settlement. That methodology potentially inflates the amount of attorney's fees in proportion to the results achieved for the class because the coupons may end up providing minimal benefit to the class. On remand, the district court should thus apply a percentage-of-redemption-value methodology for the coupon portion of a settlement, and use a lodestar method for the non-coupon part of the relief. Alternatively, the district court may use a lodestar-only methodology, but only if it does not consider the coupon relief or takes into account its redemption value.

BACKGROUND

A. Plaintiffs sue Whirlpool for allegedly faulty dishwashers.

This case began a long time ago in a district far, far away from the Central District of California: in the Maryland home of Steve Chambers and his wife, their 2002 KitchenAid dishwasher unit suffered from a bad electronic control board ("ECB") that caused it to overheat and even emit internal flames. Chambers complained to Whirlpool Corp., which makes dishwashers under its own brand name as well as under the Kenmore and KitchenAid imprints. After his requests went unheeded, he set up a website that attracted similar grievances from other Whirlpool dishwasher owners.

Ultimately, Chambers and his wife, along with eight other plaintiffs, filed a putative class action lawsuit in California against Whirlpool, asserting breach of warranty and other state law claims. The complaint alleged that several of Whirlpool's dishwashers suffered from a design defect that caused a small number of ECBs to overheat and malfunction. While the complaint highlighted the potential risk of a dishwasher malfunction and even a fire, actual instances of failure appear to be relatively rare. Apparently, fewer than 0.2% of the dishwashers have suffered overheating problems.

The initial complaint sought certification of a nationwide class of dishwasher purchasers. The plaintiffs amended their complaint four times. In the process, they added two federal claims for alleged violations of the Magnuson-Moss Warranty Act, while narrowing the scope of the class allegations to 11 state classes.

B. The parties settle — with coupons comprising most of the benefits.

The district court had not yet ruled on any substantive motion when the parties reached a nationwide settlement in September 2015. The settlement agreement provided benefits to both class and non-class members based on the type of ECB in the consumer's dishwasher.

The proposed class included people who bought dishwashers that used a "Rushmore" or "Rush" ECB manufactured between October 2000 and January 2006. The non-class dishwashers contained a "NewGen" or "Raptor" ECB manufactured between February 1998 and March 2012. The settlement covers about 5.8 million Rushmore/Rush class members and 12.6 million NewGen/Raptor non-class members.

The settlement provides overlapping benefits on a claims-made basis to both class and non-class members — with the difference being that class members are entitled to more coupons. This difference stems from the somewhat higher risk of overheating in Rushmore/Rush ECB dishwashers. Under the settlement, class members receive benefits and release potential claims unless they timely request exclusion. Non-class members, on the other hand, must separately execute a release to receive a benefit. Neither class nor non-class members release claims for personal injury or damage to property other than the dishwasher.

The settlement provides to both class and non-class members: (i) full reimbursement or $200 for individuals who paid for an overheating-related repair; (ii) $200 or $300 for consumers who replaced an overheated dishwasher; and (iii) $100 or a 30% discount "rebate" (i.e. , coupon) for a new Whirlpool dishwasher if there is a future overheating incident within two years of the settlement notice date, or within 10 years of purchase for NewGen/Raptor owners.

Class members also receive a 10–20% "rebate" coupon to purchase a new Whirlpool dishwasher, which expires 120 days after the claim deadline. And finally, Whirlpool must revise its service kit pointers and training bulletins to "emphasize the important safety function" of the thermal cut-off device that helps prevent overheating, "instruct technicians and customers not to bypass or disable" the device, and urge "inspect[ion]" of the device when servicing an ECB.

C. The district court approves the class settlement.

The district court granted preliminary settlement approval and class certification. Direct mail notice was sent to 3,567,542 class members. By the deadline, 133,040 claims had been filed (i.e. , a 3.7% response rate based on the number of direct mail notices sent). Of these, 122,294 claimed a rebate/coupon, 26,380 claimed cash reimbursement (overlapping with the rebate group), and 329 were uncategorized. The face value of the 26,380 claims submitted for cash reimbursement was $10.89 million, but only 5,249 of those claims included potentially adequate documentation (which would require further review).

Put another way, only 4% of the 133,040 filed claims — at most — could potentially involve cash reimbursement. The remaining 96% (or more) of the claims are for "rebates," i.e. , discount coupons that customers may use to buy a new Whirlpool dishwasher. In a stark sign that the parties could not agree on how many people would redeem the coupons, the parties’ valuation of the settlement diverged dramatically: Whirlpool estimated it to be as low as $4.2 million, while the plaintiffs put the high end at $116.7 million.

In October 2016, the district court granted final settlement approval and certified the settlement class. Evaluating the relevant factors, the court concluded that the settlement was "fair, reasonable, and adequate" under Rule 23(e)(2). The court also determined that the notice requirements had been met, and that all of the class certification requirements remained satisfied.

D. The district court approves nearly $15 million in attorney's fees.

The parties agreed that the attorney's fees should not come from any funds allocated for class members, but that Whirlpool would instead directly pay the plaintiffs’ class counsel. They could not agree, however, on the amount of legal fees. So the parties took a gamble and agreed that the court should decide the attorney's fees amount. The settlement agreement stipulated that any "issues relating to attorneys’ fees and costs" were to be considered separately from "the fairness, reasonableness, and adequacy" of the settlement. It also stated that, even if the district court declined to award attorney's fees in whole or in part, "the remaining provisions" of the settlement would "remain in full force and effect."

The gamble paid off handsomely for the plaintiffs’ counsel. They applied for a fee award of $15 million, based on an asserted baseline lodestar of $8,948,487.98 and a requested 1.68 upward multiplier. To support their request, the plaintiffs’ counsel claimed a settlement value in the range of $55.7 million to $116.7 million. Whirlpool countered that the fee request dwarfed the actual settlement value, which Whirlpool estimated to be between $4.2 and $6.8 million. Whirlpool sought a reduction in the baseline lodestar because, among other things, more than $2.6 million of the total was for over 8,200 hours of document review. Whirlpool further argued for a 50% negative multiplier to reasonably align the fee award with the settlement value.

The district court granted the bulk of the plaintiffs’ fee request, reducing only $130,038.75 from two lawyers’ time for billing in quarter-hour increments for simple tasks. Using a resulting lodestar of $8,818,449.23, the court applied the requested 1.68 multiplier, for a total fee award of $14,814,994.70. Noting the widely divergent settlement valuations offered by the parties, as well as the undetermined total number of valid claims to be made under the settlement, the court declined to perform a cross-check of the fee award to see if it was reasonable.

Whirlpool appeals the district court's decision to award $14.8 million in attorney's fees, but it does not challenge the court's approval of the class settlement. Objectors W. Allen McDonald, Jan Miorelli and Christine Knott appeal both the fee award and the decision granting final settlement approval.

DISCUSSION

I. Standard of Review

We review "a district court's award of fees and costs to class counsel, and its method of calculation, for abuse of discretion." In re HP Inkjet Printer Litig. , 716 F.3d 1173, 1177 (9th Cir. 2013). We review de novo the "legal bases" of a fee award. Cmty. Ass'n for Restoration of the Env't v. Henry Bosma Dairy , 305 F.3d 943, 956 (9th Cir. 2002). We review the approval of a class settlement for a "strong showing" that "the district court clearly abused its discretion." In re Hyundai & Kia Fuel Econ. Litig. , 926 F.3d 539, 556 (9th Cir. 2019) (quotation omitted). II. CAFA's attorney's fees provisions apply to all federal class actions.

To begin, we address whether the attorney's fees provisions in the Class Action Fairness Act (CAFA) even apply to a class action case based on diversity jurisdiction. If they do not, then the district court correctly applied a lodestar methodology for fees under California state law, according to the plaintiffs. The plaintiffs’ argument lacks merit.

In 2005, Congress enacted CAFA to provide a federal forum for class action lawsuits in light of a perceived "bias" in state courts and to curb alleged "abuses" in class action litigation. See Public Law 109-2, Sec. 2(a). CAFA thus requires only minimal diversity (instead of complete diversity) to assert federal jurisdiction over a class action with an amount-in-controversy exceeding $5 million. See 28 U.S.C § 1332(d)(2).

Importantly here, CAFA also established specific rules to govern fee awards for coupon settlements in federal class actions. See 28 U.S.C § 1712. It states that the attorney's fees provisions apply to any "class action," which is defined as "any civil action filed in a district court of the United States under rule 23 of the Federal Rules of Civil Procedure or any civil action that is removed to a district court of the United States that was originally filed under a State statute." 28 U.S.C. §§ 1711(2), 1712. The plain language of CAFA makes clear that its attorney's fees provisions preempt any corresponding state law and apply to any class action case in federal court, including those based on diversity jurisdiction. Cf. Hubbard v. SoBreck, LLC , 554 F.3d 742, 744 (9th Cir. 2009) (federal disability law's attorney's fees provision preempts state law). Indeed, it would be highly incongruous for Congress to expand federal jurisdiction for class action lawsuits based on diversity jurisdiction, but then in the same statute prevent CAFA's attorney's fee provisions from applying in those diversity jurisdiction-based cases.

The plaintiffs also argue that the Rules Enabling Act precludes CAFA preemption of state law on attorney's fees. But the Rules Enabling Act limits only rules prescribed by the Supreme Court, not legislation passed by Congress. See 28 U.S.C. § 2072.

Finally, the plaintiffs contend that California law governs fee calculations because the settlement agreement has a choice-of-law provision stating that "the rights and obligations of the Parties shall be construed and enforced in accordance with, and governed by, the laws of the State of California." But in Murphy v. DirecTV, Inc. , we held that parties cannot use a choice-of-law provision to opt into a California law that federal law has preempted. 724 F.3d 1218, 1225–28 (9th Cir. 2013). Under preemption, "the [preempted] rule is not, and indeed never was, California law" since "state law is nullified to the extent that it actually conflicts with federal law." Id . at 1226 (emphasis in original) (citation omitted). Thus, the choice-of-law provision here could not have invoked a California rule permitting a lodestar-only calculation because CAFA has supplanted it. See id .

III. The district court improperly used a lodestar-only method to calculate attorney's fees for the coupon portion of the settlement.

CAFA sets forth two distinct methods to calculate the award of attorney's fees in class actions: (1) the lodestar method based on the reasonable number of billable hours spent on the case ( § 1712(b) ), and (2) the percentage-of-value method in which the attorney receives a percentage of the value of the settlement ( § 1712(a) ).

This court in HP Inkjet held that CAFA mandates the use of a percentage-of-value calculation for any "portion" of a fee award "attributable to the award of the coupons." 716 F.3d at 1180–81 (quoting 28 U.S.C § 1712(a) ). We clarified in In re Easysaver Rewards Litigation that the lodestar methodology may be used in a "mixed" settlement involving coupon and non-coupon relief only if the lodestar calculation does not consider the coupon portion of the settlement or takes into account the coupon redemption value. 906 F.3d 747, 759 (9th Cir. 2018).

The district court, however, arrived at its $14.8 million fee award based solely on a lodestar valuation that considered the work performed for the coupon portion of the settlement. We therefore vacate the district court's fee award and remand.

A. Attack of the coupons: coupon settlements may endanger class members’ interests.

Federal courts, as well as Congress, have long viewed coupon-based class action settlements with a skeptical eye. And for good reason: counsel for plaintiffs and defendants have sometimes conspired to craft class action settlements that benefit themselves at the expense of the class members.

Defendants sometimes favor coupon settlements because they do not require the payment of cash out of pocket, but instead offer coupons for the company's product or service. Moreover, coupon settlements often impose onerous obstacles that make it difficult to redeem the coupons. Many plaintiffs’ counsel also prefer coupon settlements to inflate the ostensible value of the settlement — and, in turn, ratchet up their request for attorney's fees. See HP Inkjet , 716 F.3d at 1177–78 (noting that coupon settlements "decoupl[e] the interests of the class and its counsel"). And too often, class members do not benefit from a coupon settlement because many of them, not surprisingly, do not want to reward the offending company by buying its product or service again, even if they receive a discount.

In enacting CAFA, Congress struck back against this perceived menace of phantom benefits in coupon settlements. The primary purpose of CAFA was "to curb perceived abuses of the class action device," such as "the coupon settlement, where defendants pay aggrieved class members in coupons or vouchers but pay class counsel in cash." Id . at 1177 (quotation and citation omitted). CAFA thus provided a new hope to "tether the value of an attorneys’ fees award to the value of the class recovery" in coupon settlements. Id . at 1178.

B. Lodestar wars: percentage-of-value, not lodestar, methodology applies to the coupon portion of settlements.

The plain language of CAFA makes clear that a court should ordinarily use the percentage-of-value, not lodestar, methodology for the portion of the settlement involving coupons. It states that if a class action settlement provides "for a recovery of coupons to a class member, the portion of any attorney's fee award to class counsel that is attributable to the award of coupons shall be based on the value to class members of the coupons that are redeemed. " 28 U.S.C. § 1712(a) (emphasis added). In other words, § 1712(a) requires fees to be calculated as a percentage of the coupon redemption value rather than the face value of coupons. That prevents class counsel from "puff[ing] the perceived value of the settlement so as to enhance" their award. HP Inkjet , 716 F.3d at 1179.

In HP Inkjet , this court held that the percentage-of-redemption-value method applies whenever a settlement "provides for coupon relief, either in whole or in part. " Id . at 1175–76 (emphasis added). If a settlement includes "in part" coupons and also non-coupon relief, then subsections (b) and (c) of 28 U.S.C. § 1712 come into play. Id . at 1183–86.

Under § 1712(c) — which governs these so-called "mixed" settlements involving both coupon and non-coupon relief — the percentage-of-redemption-value method still applies to the portion of fees attributable to coupons. 28 U.S.C. § 1712(c)(1) (stating that fees "based upon a portion of the recovery of the coupons shall be calculated in accordance with subsection (a)," which sets forth the percentage-of-redemption-value methodology). But the remaining portion of fees attributable to "non-coupon relief" is calculated under § 1712(b) as a reasonable lodestar amount plus or minus any appropriate multiplier. 28 U.S.C. § 1712(b) (stating that where "a portion of the recovery of the coupons is not used to determine the attorney's fees," it "shall be based upon the amount of time class counsel reasonably expended"). In short, the total fee award for "mixed" settlements under § 1712(c) is the sum of: (i) "a reasonable contingency fee based on the actual redemption value of the coupons" ( § 1712(a) ); and (ii) "a reasonable lodestar amount to compensate class counsel for any non-coupon relief obtained" ( § 1712(b) ). HP Inkjet , 716 F.3d at 1184–85.

More recently, this court in In re Easysaver Rewards Litigation explained that a district court in "mixed" settlements may nonetheless opt to use the "lodestar approach provided that it does so without reference to the dollar value of [the coupon relief]" or "if it accounts for the redemption rate of the coupons in calculating the dollar value." 906 F.3d at 759. In that case, the mixed settlement involved a $3.5 million cash fund to refund members’ enrollment fees, and a coupon component providing a $20 credit to buy additional products. See id. at 752. The district court applied a multiplier to the lodestar based in part on a settlement valuation that included coupon relief. Id. at 759. We thus held that the "value of the coupon relief therefore impermissibly informed the district court's approval of the lodestar fee." Id. at 759–60.

As explained below, the district court erred by applying a lodestar-only methodology to calculate the fees, even though potentially unredeemed coupons represent most of the settlement value.

1. The parties’ settlement is a coupon settlement.

First, the plaintiffs argue that CAFA does not even apply here because the settlement is purportedly not a "coupon settlement." The plaintiffs are wrong.

CAFA itself does not define "coupon," but this court has established three factors to determine whether a settlement is a coupon settlement: (i) "whether class members have ‘to hand over more of their own money before they can take advantage of’ a credit"; (ii) "whether the credit is valid only ‘for select products or services’ "; and (iii) "how much flexibility the credit provides, including whether it expires or is freely transferrable." Easysaver , 906 F.3d at 755 (citing In re Online DVD-Rental Antitrust Litig. , 779 F.3d 934, 951 (9th Cir. 2015) ).

In Easysaver , this court applied this test to a class action settlement that included a $20 credit for use on the defendants’ websites. Id . at 752–53. We concluded that the credit was a CAFA "coupon" because: (i) the defendants offered only 15 to 25 products under $20, and shipping charges would likely bring the cost of any purchase over $20; (ii) class members could use the credits to buy a product only from the defendants; and (iii) the credits expired in one year and were subject to certain blackout periods. Id . at 756–58. We distinguished the $12 Walmart gift cards at issue in Online DVD-Rental because those gift cards never expired and could be swapped for cash, or alternatively they allowed class members to buy a wide range of low-cost products under $12. Id . at 755–58.

Here, the relevant factors establish even more persuasively than in Easysaver that the 10–20% dishwasher "rebate" is a "coupon" under CAFA, despite the settlement agreement's refusal to use that term. First, to use the "rebate," class members must spend hundreds of out-of-pocket dollars to purchase a new dishwasher. Second, the rebate applies only to Whirlpool, Kenmore, or KitchenAid-brand dishwashers — the very brands that allegedly contained the overheating defect. And finally, the rebates expire in 120 days, a third of the useful life of the Easysaver credits. Id . at 757. Given that a dishwasher typically lasts at least several years, most consumers likely will not redeem their coupons within 120 days.

2. The court erred in applying a lodestar-only methodology.

We vacate the fee award because the district court failed to follow CAFA's mandate to use a percentage-of-value calculation for any "portion" of a fee award "attributable to the award of the coupons." See 28 U.S.C § 1712(a). Nor did it use a lodestar methodology completely divorced from the coupon portion of the settlement, as permitted under Easysaver .

The district court reasoned that "CAFA authorizes the court to calculate attorney's fees utilizing the lodestar method" if "the settlement includes both coupon and monetary relief." But we foreclosed that argument in HP Inkjet when we held that the percentage-of-redemption-value method applies whenever a settlement "provides for coupon relief, either in whole or in part. " 716 F.3d at 1175–76 (emphasis added). The district court also erred in holding that CAFA's mixed settlement provision ( 28 U.S.C § 1712(c) ) does not apply because it does "not contemplate ... settlements that involve coupon relief and monetary relief ." While mixed settlements "that award coupons and monetary relief are not expressly mentioned in In re HP ," it would relegate such settlements to "a no-man's land" if they are not included within the scope of § 1712(c). Easysaver , 906 F.3d at 759 n.11 (applying § 1712(c) for a settlement that included a $3.5 million fund and coupon relief). We thus held that in such mixed settlements, a district court may "use the lodestar approach" only "to determine any portion of attorney's fees not attributable to coupons." Id . at 759. Because we issued Easysaver after the district court had already determined its award, it understandably did not take into account that holding.

The district court started off with an $8.8 million lodestar based on the claimed hourly fees of class counsel, except for a small $130,038.75 reduction for two lawyers’ time. That amount encompassed all of the work performed by the plaintiffs’ counsel for the entire case, and thus necessarily included the work completed on behalf of the coupon portion of the settlement. And then the court applied a 1.68 multiplier, again taking the coupons into account. See Chambers v. Whirlpool Corp. , 214 F. Supp. 3d 877, 902 (C.D. Cal. 2016) ("impressive" result includes "insurance-like coverage for future Overheating Events," i.e. , a rebate/coupon option for new dishwasher if it overheats); id . ("create[s] an incentive for current owners to replace their Class Dishwashers" by using coupons); id . at 904–05 (acknowledging that settlement value may be as high as $116.7 million in refusing to cross-check fees). Because the lodestar amount and the multiplier implicitly and explicitly took into account the coupon portion of the settlement, the district court's fee award conflicts with both HP Inkjet and Easysaver.

3. The court must separately consider the coupon portion of the settlement on remand.

On remand, the district court should first attempt to ascertain the (a) the redemption value of the coupons ( § 1712(a) ), and (b) the value of the non-coupon portion of the settlement ( § 1712(b) ). See HP Inkjet , 716 F.3d at 1184–85.

a. Percentage-of-redemption value of the coupon portion of settlement — § 1712(a).

The first component of the fee assessment under § 1712(a) is to calculate the coupon redemption value of the settlement (i.e. , the collective value of the coupons redeemed by class members). CAFA appears to have contemplated coupon settlements in which coupons are redeemed before final settlement approval. But as in HP Inkjet , the parties here "essentially invited the error" by structuring a settlement where the coupon redemption value was unknown at the time of final approval. 716 F.3d at 1186.

Nevertheless, courts are equipped with options — such as a "bifurcated or staggered" fee award — to address such a scenario. See id . at 1186 n.19. Bifurcation is appropriate here, where the class members have a relatively short 120-day redemption window to decide whether to redeem their coupons to buy a new Whirlpool dishwasher. Once the collective redemption value is known after 120 days, the district court will be able to calculate a "reasonable contingency fee" based on that coupon redemption value amount. See id . at 1184.

In setting the "reasonable contingency" fee for the coupon portion of the settlement, the district court should give a hard look at the resulting fee amount to ensure that it is proportional to the coupon benefits provided to the class. Cf. In re Bluetooth Headset Prods. Liab. Litig. , 654 F.3d 935, 941 (9th Cir. 2011) (courts have an "independent obligation to ensure that the [fee] award ... is reasonable."); Hensley v. Eckerhart , 461 U.S. 424, 440, 103 S.Ct. 1933, 76 L.Ed.2d 40 (1983) ("[T]he district court should award only that amount of fees that is reasonable in relation to the results obtained."). The potential danger that class counsel may elevate their own interests over the class members’ always lurks in class settlements. Cf. Richard A. Posner, Economic Analysis of Law, at 627 (Aspen 5th ed 1998) ("lawyer for the class will be tempted to offer to settle with the defendant for a small judgment and a large legal fee, and such an offer will be attractive to the defendant, provided the sum of the two figures is less than the defendant's net expected loss from going to trial").

We recognize that some settlements may involve coupons that do not expire for many years, or never expire at all. Here, the settlement also provides continuing coverage to NewGen/Raptor dishwasher owners for future overheating events, which extends into 2021. Those who experience a future overheating event are entitled to either $100 or a 30% dishwasher coupon. For coupons with extended redemption periods, courts may use a "staggered" method in which the plaintiffs’ counsel is paid periodically as the coupons are redeemed. See HP Inkjet , 716 F.3d at 1186 n.19.

b. Lodestar value of non-coupon portion of settlement — § 1712(b).

This leads us to the second component of the fee assessment under § 1712(b) : a "reasonable lodestar amount to compensate class counsel for any non-coupon relief obtained." Id . at 1185. Our court does not appear to have articulated how to determine a "reasonable lodestar amount" for "non-coupon relief." We can think of at least two ways to do so:

First, a court can take the lodestar for the entire case, and then use a negative multiplier to discount that sum because the lodestar, by definition, includes work done on behalf of the coupon portion of the settlement. In determining the negative multiplier, the court can, for example, seek to assess what portion of the settlement value stems from coupon relief.

Second, a court can try to parse through the billable hours to apportion which hours were reasonably expended for the coupon portion of the settlement and which for the non-coupon portion. We acknowledge that there may not be clear lines delineating between the two portions, but courts engage in similar tasks to determine attorney's fees in other contexts. Cf. , e.g. , Schwarz v. Sec'y of Health & Human Servs. , 73 F.3d 895, 904 (9th Cir. 1995) ("Once a district court concludes that a plaintiff has pursued unsuccessful claims that are unrelated to the successful claim, its task is to exclude from the calculation of a reasonable fee all hours spent litigating the unsuccessful claims.").

Regardless of which method a court uses to establish the lodestar amount for the non-coupon portion of the settlement, it should apply a lodestar "cross-check" in mixed settlements, or in the alternative, articulate why it is not feasible in a particular case. While we do not ordinarily require a lodestar "cross-check," In re Hyundai , 926 F.3d at 571, the analysis is distinct for mixed settlements involving both coupon and non-coupon relief. This is so because the lodestar represents the presumptive value of class counsel's work for the entire case, and therefore risks double-counting fees for work attributable to the coupon portion of a mixed settlement. And even if a court adopts the second method of pruning the lodestar of coupon-related legal work, a cross-check would provide assurance that the court accurately divvied up the legal work.

Thus, to ensure the § 1712(b) lodestar calculation does not overcompensate class counsel for work unrelated to non-coupon relief, the district court must ascertain the value of the non-coupon portion of the settlement. We have previously held that where "the lodestar amount overcompensates the attorneys according to the 25% benchmark standard, then a second look to evaluate the reasonableness of the hours worked and rates claimed is appropriate." In re Bluetooth , 654 F.3d at 945 (quoting In re Coordinated Pretrial Proceedings , 109 F.3d 602, 607 (9th Cir. 1997) ). This principle is particularly apt in the context of § 1712(b) — for example, if the lodestar exceeds the non-coupon value of a settlement, it stands to reason that the excess amount likely includes fees attributable to coupon relief.

This is well-illustrated by the vastly contrasting values that the parties attribute to the non-coupon portion of the settlement: about $3 million according to Whirlpool, compared to the $63 million figure advanced by the plaintiffs. If Whirlpool is correct that the non-coupon relief is worth at most around $3 million, then the baseline lodestar amount of $8,818,449 calculated by the district court would be clearly disproportionate. That would merit a significant negative multiplier. On the other hand, a lodestar of $8,818,449 in relation to a non-coupon value of $63 million appears reasonable.

The district court held that the massive gap between the parties’ dueling valuations signaled that any attempt to value the settlement "would be imprecise to the point of uselessness." But it becomes even more critical to cross-check the lodestar valuation if the parties present widely divergent settlement valuation estimates. It may admittedly be difficult to determine that amount with precision, but courts must try to do so to ensure the fees are not excessive.

In any event, the record provides ample evidence, including useful expert analysis, to enable the court to apply reasonable assumptions to estimate the overall non-coupon value. The court can evaluate evidence on the likely deficiency rate of the dishwashers, the value of coverage for future deficiencies, and other relevant factors.

Based on our review of the record, it appears that the large gap between the settlement valuation estimates can be narrowed significantly. For instance, the plaintiffs use the $10,892,286 "face value" figure stated in the settlement administrator's declaration to determine the value of cash reimbursements. But they ignore the declaration's accompanying explanation that over 80% of those claims were deemed to be facially deficient, and that the remainder had yet to be evaluated for sufficiency of documentation.

Similarly, the plaintiffs’ assertion of a $50 million value for future overheating coverage assumes 28,000 overheating claims per year until the end of coverage in 2021. That projected annual figure, however, outstrips the total overheating claims over the decade before, and the record suggests that the rate of such events should decline over time. Conversely, Whirlpool appears to assume that the extended warranty has no value if a dishwasher never suffers an overheating failure, but perhaps that limited "insurance-like" coverage has some monetary value. Expert evidence and careful assessment of the record can aid the court.

The district court also noted that the enhanced safety warnings required by the settlement "cannot be quantified with precision, if at all." We agree that courts need not try to attach a precise dollar figure to these types of non-monetary relief. But the district court can, based on the record, determine the significance of this benefit, and employ it as a qualitative factor in deciding whether a multiplier is warranted.

The net result of this analysis is that the total fee award for "mixed" settlements, as set forth in § 1712(c), will be the sum of the two calculations under (i) § 1712(a), i.e. , a reasonable contingency percentage of the total coupon redemption value; plus (ii) § 1712(b), i.e. , class counsel's reasonable lodestar, cross-checked against the value of the non-coupon relief, and adjusted based on any applicable multiplier factors. See § 1712(c) (setting forth "[a]ttorney's fees award calculated on a mixed basis in coupon settlements"); In re HP Inkjet , 716 F.3d at 1185 ("the total amount of fees awarded under subsection (c) will be the sum of the amounts calculated under subsections (a) and (b)").

In most cases, we expect district courts to add the sums under § 1712(a) and § 1712(b) to arrive at the attorney's fees in "mixed" settlements, as set forth in § 1712(c). But in EasySaver we noted that a district court may still award fees based solely on a lodestar methodology if (1) "it does so without reference to the dollar value of [the coupon relief]" or (2) "if it accounts for the redemption rate of the coupons in calculating the dollar value." 906 F.3d at 759.

For the first EasySaver option — lodestar without reference to coupon relief — it will effectively be the amount allowed under § 1712(b), i.e. , class counsel's reasonable lodestar, cross-checked against the value of the non-coupon relief, and adjusted based on any multiplier factors. This option potentially shortchanges plaintiffs’ counsel by omitting the fees due under § 1712(a), so this option should be used only if there is no reasonable way to calculate the reasonable contingency percentage of the coupon redemption rate.

For the second EasySaver option — lodestar that takes into account the coupon redemption rate — a district court would start off with the lodestar for the entire case, and then apply a negative or positive multiplier if warranted. We believe that a district court should ordinarily try to calculate fees for "mixed" settlements by following § 1712(c) ’s methodology of adding the sums under §§ 1712(a) and 1712(b). It should choose this second EasySaver option only if it becomes too difficult to calculate the fees under § 1712(c), and it should provide an explanation for choosing this second EasySaver option.

IV. The court erred in awarding a 1.68 lodestar multiplier.

Because of a "strong presumption that the lodestar is sufficient," a multiplier is warranted only in "rare and exceptional circumstances." Perdue v. Kenny A. ex rel. Winn , 559 U.S. 542, 546–52, 130 S.Ct. 1662, 176 L.Ed.2d 494 (2010) (quotation omitted). A multiplier "may not be awarded based on a factor that is subsumed in the lodestar calculation." Id . at 553, 130 S.Ct. 1662 ; see also In re Bluetooth , 654 F.3d at 942 n.7 (many of the Kerr factors for determining reasonable attorney's fees are "subsumed within the initial calculation of hours reasonably expended at a reasonable rate") (citation omitted); Parsons v. Ryan , 949 F.3d 443, 467 (9th Cir. 2020) ("Any reliance on factors that have been held to be subsumed in the lodestar determination will be considered an abuse of the trial court's discretion.") (citation omitted). Moreover, a multiplier must be "supported by both ‘specific evidence’ on the record and detailed findings by the lower courts that the lodestar amount is unreasonably low or unreasonably high." Van Gerwen v. Guarantee Mut. Life Co. , 214 F.3d 1041, 1045 (9th Cir. 2000) (citations omitted).

The district court here granted the plaintiffs’ request for a 1.68 multiplier because: (i) litigating the case "required an extraordinary amount of time and labor"; (ii) there were "difficult and complex" legal questions; (iii) the case was "undesirable" because of "substantial litigation risks" and the fact that the defendants are "large corporations with substantial resources"; (iv) the settlement results are "impressive"; and (v) attorney's fees hinged on success.

To begin with, the district court incorrectly included the value of the coupon portion of the settlement in establishing the 1.68 multiplier for the lodestar value. But as discussed above, the lodestar — or the multiplier — cannot reflect the work done for the coupon portion of the relief. Further, the reasons cited by the district court cannot justify enhancement and are not tied to the multiplier amount.

A. Time and Labor

We calculate the baseline lodestar figure "by multiplying the number of hours reasonably expended on the litigation by a reasonable hourly rate." Id . In the fee motion, class counsel submitted their total number of hours spent litigating this case, along with proffered hourly rates. Apart from a minor reduction of two lawyers’ time, the district court accepted class counsel's lodestar submission. Thus, because the time and labor spent by class counsel were "subsumed" within this lodestar figure, they could not justify an upward multiplier. See Merritt v. Mackey , 932 F.2d 1317, 1324 (9th Cir. 1991) ("The time involved is clearly subsumed in the lodestar figure.").

Even if a court could consider time and labor in some cases, it appears questionable to do so here because of the staggering number of written discovery hours — including over $2.6 million dollars in fees for document review — without considering the asymmetrical nature of discovery in class actions that can lead to excessive billing. In most complex litigation matters, parties face a "mutually assured destruction" scenario that theoretically curbs excessive discovery demands: if one party propounds burdensome discovery requests, the other side is likely to respond in kind. See Jorling v. Anthem, Inc. , 836 F. Supp. 2d 821, 830 n.5 (S.D. Ind. 2011) ("[I]f two similarly sized entities are litigating, the discovery process is more reciprocal, meaning parties have the incentive to reach reasonable agreements on the scope of discovery. In other words, if they don't live by the Golden Rule — ‘Do unto others as you would have them do unto you’ — they could both face onerous discovery costs.").

By contrast, class action plaintiffs typically possess no or very limited discoverable materials, while defendants may have reams of documents and terabytes of electronic data. Class action plaintiffs thus have an incentive to seek aggressive discovery (and log a tremendous number of hours in the process) without fear of reciprocally burdensome discovery. Cf. id. ("The discovery process in securities cases has often been described as ‘asymmetrical’ because the defendant has a large universe of documents that are of interest to the plaintiff, whereas the plaintiff has relatively few documents of interest to the defendant. Therefore, the plaintiff has the incentive to pry into every imaginable crevice of the defendant's records, thus forcing the defendant to incur substantial costs or settle."); Boeynaems v. LA Fitness Int'l, LLC , 285 F.R.D. 331, 334–35 (E.D. Pa. 2012) (allocating discovery costs to plaintiffs due to "asymmetrical discovery" where the class had "very few documents" while the defendant had "millions of documents and millions of items of electronically stored information").

This is not to suggest that class action plaintiffs engage in unnecessary or excessive discovery. Rather, when considering whether an upward multiplier should apply to a large lodestar based on a high number of discovery-related hours, courts should assess the reasonableness of the discovery efforts in light of the lack of structural restraints on discovery in class action cases. The district court did not do so here, and thus the enhancement based on time and labor is improper for this reason as well.

B. Difficult and Complex Legal Questions

The district court also considered that the "case involved a number of difficult and complex legal questions." But "the novelty and complexity of a case generally may not be used as a ground for an enhancement because these factors ‘presumably are fully reflected in the number of billable hours recorded by counsel.’ " Perdue , 559 U.S. at 553, 130 S.Ct. 1662 (quoting Blum v. Stenson , 465 U.S. 886, 889, 104 S.Ct. 1541, 79 L.Ed.2d 891 (1984) ); see also Greater Los Angeles Council on Deafness v. Cmty. Television of S. Cal. , 813 F.2d 217, 221 (9th Cir. 1987) ("[T]he novelty and difficulty of issues are inappropriate factors to use in enhancing a fee award[.]").

In certain contexts — such as where a subset of class counsel has borne a disproportionate brunt of the difficult aspects of a case — we have recognized that "complexity" can warrant enhancement. See Hyundai & Kia , 926 F.3d at 571–72 (downward multipliers applied to another subset of class counsel). The district court here did not identify any exceptional factor, and instead issued a blanket multiplier to all class counsel based partly on the general presence of "difficult and complex legal questions." In fact, far from being complex, an enormous amount of class counsel's work involved routine tasks such as document review. Enhancement on this basis was therefore improper. C. Undesirability of Case

The district court considered this case "undesirable" because of: (i) difficulties the plaintiffs may have encountered recovering on their claims; and (ii) the fact that the defendants are deep-pocketed corporations. But in City of Burlington v. Dague , the Supreme Court explained that, because of the perverse incentives it would create, a multiplier should not be based on the legal or factual merits of a claim. 505 U.S. 557, 562–63, 112 S.Ct. 2638, 120 L.Ed.2d 449 (1992). The Supreme Court also noted that the lodestar typically reflects the difficulty of establishing the merits of a claim. See id . at 562, 112 S.Ct. 2638.

The record also offers no meaningful evidence that the defendants’ resources made this an undesirable case to pursue. If the mere fact that the defendants are "large corporations" were sufficient, then most class action fee awards would automatically qualify for enhancement — contrary to the rule that multipliers are for "rare and exceptional circumstances." See Perdue , 559 U.S. at 552, 130 S.Ct. 1662. In practice, deep pockets often create an incentive to sue, particularly in the class action context. And indeed, the record here undercuts the notion of undesirability, as five different law firms pursued these claims against Whirlpool and collectively litigated this case for many years.

D. Impressive Results

The district court lauded the settlement as "impressive" based on the relief secured and the fact that "class counsel began with an 11-state lawsuit and converted it into a nationwide settlement." But in Hensley v. Eckerhart , the Supreme Court determined that a success-based multiplier could not be awarded simply because of a trial court's finding that the relief secured "was substantial" or "of significant import." 461 U.S. at 438, 103 S.Ct. 1933. The Court required greater specificity in "the relationship between the amount of the fee awarded and the results obtained." Id . at 437, 103 S.Ct. 1933. The Court reasoned that a fee reduction would be warranted if "the relief, however significant, is limited in comparison to the scope of the litigation as a whole." Id . at 438–40, 103 S.Ct. 1933.

Here, the district court's findings did not adequately support a success-based multiplier because it neither assessed the actual value of the settlement, nor compared it to "the scope of the litigation as a whole." Id . at 440, 103 S.Ct. 1933. While observing that the parties’ respective valuations of the settlement ranged from $4,220,000 to $116,700,000, the court declined to determine where in that spectrum the actual value fell. Given this enormous spread, without at least estimating the settlement value, the court could not have conducted the necessary evaluation between "the extent of success and the amount of the fee award." Id . at 438, 103 S.Ct. 1933. Indeed, only 4% of the submitted claims involved cash reimbursements; the remaining claims are for coupons, and the parties diverge dramatically on the valuation of those coupons. Without determining their value, it is difficult to assess whether the results achieved are "impressive."

We also disagree with the district court's conclusion that the nationwide scope of the settlement made it "particularly impressive." First, class counsel did not "beg[i]n with an 11-state lawsuit." Rather, the initial complaint sought certification of a nationwide class, and the plaintiffs later narrowed the class to 11 states after multiple amendments in response to Whirlpool's motions to dismiss. Second, class action defendants — if they decide to settle — often prefer a nationwide settlement to secure the broadest release possible. Not surprisingly, Whirlpool states that it "demanded" a nationwide release "to prevent copycat litigation." And finally, without determining the settlement value, the district court had an insufficient basis to conclude that the nationwide scope reflected favorably on class counsel's efforts.

E. Contingency Risk

The district court considered the "contingent nature of success" to be an "extremely important factor" in warranting lodestar enhancement. We disagree.

In Dague , the Supreme Court held that "enhancement for contingency is not permitted" in certain statutory fee-shifting cases. 505 U.S. at 567, 112 S.Ct. 2638. The Court recognized that the rationale underlying a potential multiplier on this basis is that a contingency attorney "pools the risks presented by his various cases," some of which lead to no compensation. Id . at 565, 112 S.Ct. 2638. The Supreme Court rejected this as a proper premise for enhancement because it would unfairly force a defendant to compensate the prevailing attorney for "cases where his client does not prevail." Id . (emphasis in original).

We determined Dague to be inapplicable to class action settlements involving a common fund. In re Wash. Pub. Power Supply Sys. Sec. Litig. , 19 F.3d 1291, 1300–01 (9th Cir. 1994) (" WPPSS "). We based that ruling on the fact that common fund cases do not pose the same concerns expressed in Dague because class counsel are paid "by members of the plaintiff class" rather than "by the losing defendant." Id . at 1300. As a result, a contingency multiplier in that context adheres to "the equitable notion that those who benefit from the creation of the fund should share the wealth with the lawyers whose skill and effort helped create it." Id . (citations omitted).

The settlement here presents a third category that neither Dague nor WPPSS addressed: a class action settlement where the attorney's fees are paid directly by the defendants rather than coming out of the class recovery. See Wing v. Asarco Inc. , 114 F.3d 986, 989 (9th Cir. 1997) (acknowledging the unresolved question of Dague ’s applicability to a non-common fund class settlement, but declining to reach the issue because the multiplier was justified on other grounds). We hold that Dague applies in this context because the distinguishing feature of WPPSS —that the class client was paying the contingency premium, see 19 F.3d at 1300 — is not present here. Any attorney's fees set by the court will be paid directly by Whirlpool, without reducing the class recovery, because there is no common fund. Permitting a general contingency multiplier here would thus invoke the same concern articulated in Dague — that Whirlpool would be subsidizing class counsel's losses in other cases. 505 U.S. at 565–67, 112 S.Ct. 2638.

F. Downward Multiplier

The "most critical factor" in determining the reasonableness of a fee award is "the degree of success obtained." See Hensley , 461 U.S. at 436, 103 S.Ct. 1933. Whirlpool asserts that a 0.5 downward multiplier is warranted here because of class counsel's lack of success.

Because the degree of success depends on the settlement value, whether a downward multiplier is warranted will depend on the district court's valuation of the settlement. We remand for the district court to make this determination in the first instance for its calculation of fees under §§ 1712(a) and (b) based on the principles outlined in this opinion.

V. The court did not abuse its discretion in approving the settlement.

The settlement provides that "[t]he Court's or an appellate court's failure to approve, in whole or in part, any award of attorneys’ fees and costs to Class Counsel, or any Service Award, shall not affect the validity or finality of the Settlement." For this reason, our review of the district court's decision to grant final settlement approval is not tied to our reversal of the fee award. See In re Bluetooth , 654 F.3d at 945 ("Approval of the settlement agreement was not conditioned on the award of attorneys’ fees and costs or an incentive award, and therefore our vacatur of the fee award does not necessitate invalidation of the approval order."); Easysaver , 906 F.3d at 763 (vacating fee award but affirming settlement approval).

While the objectors raise various challenges to the settlement, none of their arguments establish a "strong showing" that the district court "clearly abused its discretion" by approving the settlement. See Hyundai & Kia , 926 F.3d at 556.

A. Fairness of Settlement

Rule 23(e)(2) provides that a court may approve a class settlement "only on finding that it is fair, reasonable, and adequate." Fed. R. Civ. P. 23(e)(2). This requires a balancing assessment of:

(1) the strength of the plaintiffs’ case; (2) the risk, expense, complexity, and likely duration of further litigation; (3) the risk of maintaining class action status throughout the trial; (4) the amount offered in settlement; (5) the extent of discovery completed and the stage of the proceedings; (6) the experience and views of counsel; (7) the presence of a governmental participant; and (8) the reaction of the class members to the proposed settlement.

Churchill Vill., L.L.C. v. Gen. Elec. , 361 F.3d 566, 575 (9th Cir. 2004). The court must also determine that the settlement is not "the product of collusion among the negotiating parties." Id . at 576.

The district court evaluated those eight factors and properly found that they each support a finding of fairness except for the government participant factor, which is inapplicable. The district court also correctly determined there was no collusion because: (i) the parties did not agree to an amount or range of attorney's fees, but left the matter to the court; (ii) there is no common fund that reverts back to Whirlpool; and (iii) the parties settled via arm's length negotiations before an experienced mediator.

Objector McDonald does not challenge any of the fairness factors or contend that the parties colluded. He argues instead that, because the settlement provides only coupon relief to over 99% of the class, the district court should not have granted approval.

Coupons, however, "may be particularly appropriate in situations ‘where they provide real benefits to consumer class members.’ " HP Inkjet , 716 F.3d at 1178 n.4 (quoting S. Rep. No. 109-14, at 31). McDonald acknowledges that only 0.17% of the class experienced a dishwasher overheating event. Viewed in this light, the settlement structure is defensible. For the 0.17% with malfunctioning products, the agreement provides for repair or replacement reimbursement. And for the 99.83% who suffered no malfunction, the settlement provides both a rebate off a new dishwasher and extended coverage for future overheating events. Coupons and extended warranty thus reasonably provide some benefits to class members with fully operational dishwashers.

B. Intra-Class Conflicts

Objector Miorelli asserts that conflicts between class and non-class members, as well as between the two subclasses (past overheating and future overheating), violate Rule 23(a). This argument, however, stems from misstatements of fact and law.

First, Miorelli claims that the settlement causes "improper comingling" by providing non-class members with "benefits in exchange for the release paid for by class members." This is a misreading of the settlement agreement, which requires non-class members to execute a separate release as consideration for receiving settlement compensation.

Second, Miorelli argues that $4,000 service payments to several non-class members violate the typicality and adequacy requirements of Rule 23(a)(3) and (4). Those provisions, however, apply to "representative parties." See Fed. R. Civ. P. 23(a)(3), (4). Because the non-class members receiving service awards are not considered class representatives under the settlement, their receipt of payments is irrelevant to the Rule 23(a) typicality and adequacy analyses.

Third, Miorelli contends that the benefits to non-class members are evidence of class counsel's self-dealing, as they inflate the overall settlement value for fee calculation purposes while according no value to the class. But as already noted, non-class compensation is independent of the class recovery, and requires non-class members to execute a separate release as consideration. If class counsel overstated the value of the non-class relief to justify fees, that issue goes only to the proper amount of class counsel's fee award.

Fourth, Miorelli relies on Ortiz v. Fibreboard Corp. , 527 U.S. 815, 119 S.Ct. 2295, 144 L.Ed.2d 715 (1999) for the proposition that holders of present and future claims require separate representation because of their adversarial positions. But the Ortiz holding was directed toward a limited fund context in which compensation for present claimholders would reduce the relief available to future claimholders. 527 U.S. at 821–59, 119 S.Ct. 2295. Under the claims-made settlement here, the availability of relief for future overheating class members is not affected by the claims of past overheating class members. Ortiz thus does not apply.

C. $100,000 Payment to Chambers

Objectors McDonald and Knotts argue that Whirlpool's purchase of named plaintiff Steven Chambers’ two websites (which complained about his overheating dishwasher) for $100,000 undermines his adequacy as a class representative. But even if we assume Chambers is an inadequate class representative, this does not affect the validity of the settlement because there are 13 other unchallenged class representatives.

In Local Joint Exec. Bd. of Culinary/Bartender Tr. Fund v. Las Vegas Sands, Inc. , we considered a challenge to the adequacy of two class representatives. 244 F.3d 1152, 1162 (9th Cir. 2001). After finding that one of the individuals was an adequate representative, we declined to address the adequacy of the second individual on the basis that "the adequacy-of-representation requirement is satisfied as long as one of the class representatives is an adequate class representative." Id . at 1162 n.2. Likewise here, the presence of 13 adequate class representatives renders moot any challenge to the adequacy of Chambers.

D. Notice Issue

McDonald maintains that an error in 7,485 of the long-form notices sent to class members — which did not update the claim, exclusion, or objection deadlines after the court granted a 25-day extension — deprived those individuals of due process. Each of those class members, however, requested a long-form notice either on the settlement website or through the interactive voice response system, both of which provided the correct dates. Thus, the error was harmless. See Online DVD-Rental , 779 F.3d at 947 (notice "not perfect" because it did not include a court update, but it was sufficient since the updated information appeared elsewhere, including the settlement website).

E. Allocation of Attorney's Fees

Knott relies on a Fifth Circuit case, In re High Sulfur Content Gasoline Products Liability Litigation , 517 F.3d 220 (5th Cir. 2008), for the proposition that the district court had to judicially allocate fees among class counsel. That case, however, addressed the due process implications of approving without scrutiny a fee-splitting arrangement at an ex parte proceeding that excluded many of the affected attorneys. High Sulfur , 517 F.3d at 230–35. No similar process took place here, and in any event, issues related to the fee award do not change the propriety of settlement approval. See Bluetooth Headset , 654 F.3d at 945.

F. Access to Court

"It is well established that district courts have inherent power to control their docket." Ready Transp., Inc. v. AAR Mfg., Inc. , 627 F.3d 402, 404 (9th Cir. 2010) (citations omitted). While "the inherent powers permit a district court to go as far as to dismiss entire actions to rein in abusive conduct," a court can also "use less drastic measures such as striking documents from the docket to address litigation conduct." Id .

The record reflects that McDonald failed to cooperate with the plaintiffs’ efforts to secure discovery that other objectors had been required to provide. After McDonald moved to quash a deposition subpoena, a Tennessee judge ordered his deposition to proceed with certain limitations. Because of the impending final approval hearing, the plaintiffs offered to accept a declaration from McDonald instead of a deposition. McDonald's decision to rebuff this reasonable accommodation provided ample justification for the district court to strike his objection. See id . at 404–05.

McDonald's complaint that he did not have electronic filing privileges also lacks merit. As he acknowledges, the district court did not restrict his ability to manually file documents.

G. Ad Hominem Statements

Miorelli contends that the district court denied her objection based on class counsel's ad hominem attacks. We agree with the general principle that parties to litigation should refrain from employing ad hominem rhetoric. See Cal. Attorney Guidelines of Civility & Prof. § 8. Here, however, the district court made clear that it "considered all of the arguments set forth by the serial objectors," and denied them on the merits. There is no indication that the district court denied the objection because of class counsel's complaints that the objectors’ lawyers had filed objections to other class settlements.

H. Sealing Billing Records

Finally, Miorelli argues that, based on our ruling in Yamada v. Nobel Biocare Holding AG , 825 F.3d 536 (9th Cir. 2016), she is entitled to full access to class counsel's billing records. But Yamada held that the defendants — who would be paying class counsel's fees — were entitled to "access to the timesheets ... so they can inspect them and present whatever objections they might have concerning the fairness and reasonableness of Plaintiffs’ fee request." 825 F.3d at 544–46. Because Whirlpool had unimpeded access to class counsel's billing records, Miorelli's argument fails. In any event, because Miorelli did not oppose sealing or raise the issue in her objection, she has waived it. See Janes v. Wal-Mart Stores Inc. , 279 F.3d 883, 887 (9th Cir. 2002).

AFFIRMED IN PART, VACATED AND REMANDED IN PART.